Nvidia on Monday unveiled its latest batch of technology focused on the areas of graphics, AI, enterprise and edge computing, robotics, and remote collaboration. The company, which is holding its virtual GTC 2020 event this week, introduced a cloud streaming video AI platform called Maxine, new Ampere GPUs for visual computing, the CloudXR on AWS platform, and the Omniverse design and collaboration platform.
The RTX A6000 and the A40 are Nvidia's latest GPU designs based on Ampere architecture. The RTX A6000 is designed for the new era of visual computing, Nvidia said. The GPU will replace the Turing version of the Quadro, while the A40 -- which is a passive cooling version of the same card -- is the successor to the RTX 6000 and RTX 8000 GPUs.
Nvidia said the GPUs -- which include 48GM of memory and support for new TF32 and BF16 data formats -- are targeted at visual compute use cases such as rendering and virtual workstations, with Nvidia AI and machine learning software running on the entire product line.
"The pandemic is driving the need for even more efficiency in computing and AI," said Bob Pette, Nvidia's VP and GM of professional visualization. "Ampere from a graphics standpoint is a perfect blend of shader, Tensor core, RT Core performance and is very power efficient."
Nvidia's Maxine cloud platform has audio, video and conversational AI software to improve video conferencing. The toolkit uses new AI-based video compression technology running on Nvidia GPUs to reduce the amount of bandwidth required for video calls. Nvidia said Maxine's video compression can reduce the bandwidth needed for calls by 90 percent compared to H.264 compression.
The platform also promises to solve video conferencing pain points around face alignment, gaze correction and background noise removal, and offers additional conversational AI services such as translations, closed captioning and transcriptions via Nvidia Jarvis. Nvidia said Maxine utilizes AI microservices running in Kubernetes, allowing users to run multiple AI features simultaneously while still adhering to application latency requirements.
"Video conferencing is now a part of everyday life," said Ian Buck, VP and GM of Accelerated Computing at Nvidia. "Nvidia Maxine integrates our most advanced video, audio and conversational AI capabilities to bring breakthrough efficiency and new capabilities to the platforms that are keeping us all connected."
Meanwhile, Nvidia also introduced the Omniverse platform, designed for collaboration and simulation specifically in the areas of design, robotics, autonomous vehicles and media and entertainment. Using the platform -- which is based on Pixar's widely adopted Universal Scene Description (USD), a format for universal interchange between 3D applications -- remote teams can collaborate simultaneously on projects just as they as easily as they would jointly edit a document online, Nvidia said.
Omniverse will enter open beta this fall. Nvidia said the open beta of follows a year-long early access program in which Ericsson, Foster + Partners, ILM and over 40 other companies evaluated the platform and provided feedback to the Nvidia engineering team.
Nvidia's CloudXR platform allows organizations to create and deliver wireless AR and VR experiences from any application based on OpenVR. The SDK runs on Nvidia servers located in the cloud or on-premises data centers, delivering the advanced graphics performance needed for wireless virtual, augmented, or mixed reality environments, the company said. The CloudXR platform includes the CloudXR SDK, the Quadro Virtual Workstation (QvWS) software and Nvidia AI SDKs to deliver photorealistic graphics.
- Avaya to integrate Nvidia's Maxine cloud streaming video AI into Avaya Spaces
- Nvidia unveils Jetson Nano 2GB for students and robot hobbyists
The platform will now be available on Amazon EC2 P3 and G4 instances, which support Nvidia V100 and T4 GPUs, allowing cloud users to stream immersive experiences to remote VR and AR devices, Nvidia said. CloudXR on AWS will be generally available early next year, with a private beta available in the coming months.